October 21, 2015
As Prepared for Delivery
Thank you so much, Simon for that introduction. I’m pleased to be with all of you today at this important conference.
I’d like to thank my colleagues at the Federal Reserve Board, the New York Federal Reserve Bank, the Securities and Exchange Commission and the Treasury Department for co-sponsoring this event, and particularly the New York Fed for hosting.
I would also like to acknowledge – and thank – staff from all these agencies and the CFTC who worked so tirelessly to put together the joint report on October 15.
Today’s conference is a great opportunity to discuss the state of the Treasury market generally. The Commodity Futures Trading Commission is responsible for overseeing the Treasury futures market, as part of our general oversight of the nation’s derivatives markets. And today, I would like to draw on that regulatory experience to discuss three things.
First, I want to discuss some of the characteristics of the Treasury futures market and in particular – the characteristics that differentiate it somewhat from the cash Treasury market. Second, I want to discuss whether events like what happened on October 15 take place in other products. And finally, I want to talk about next steps, including some specific actions we may take in the next few months. These actions are not in response to October 15, but they do address issues related to market structure and market risk that have received greater attention in light of what happened on that day.
Today’s Futures Market
So first, let me discuss a few aspects of the Treasury futures market that I think are important to keep in mind as we look at these issues. The joint staff report illustrates the high correlation between prices and activity in the Treasury futures and Treasury cash markets. That may lead us to consider them as two expressions of the same – or similar – trading interests. However, there are some different market structure elements. These are probably familiar to many of you – but they are worth noting at the outset.
The futures market is primarily a centralized market. Treasury futures contracts are listed on the Chicago Board of Trade, a regulated futures exchange. Actual trading occurs on the GLOBEX electronic trading system operated by its parent, the CME Group. Under the law and Commission rules, these and all other futures contracts can only be traded pursuant to the rules of the exchange on which they are listed. The vast majority of trades take place on the electronic trading platform. There are exceptions for executions like block trades, which in the case of Treasury futures represented less than one percent of the total volume in 2014.
Certain markets, such as cash Treasuries and swaps, have a dual structure – a multilateral, exchange-style wholesale market, and a dealer-to-customer market. In contrast, all orders in the futures markets get sent to a centralized order book. Electronic trading has democratized access. Anyone can choose to be a liquidity provider. And anyone can choose to trade against an order sitting at the top of the book. All trades are reported, on a post-trade basis, and similar to the stock market ticker are disseminated widely via several outlets.
Given these characteristics, it is perhaps not surprising that electronic trading came to the futures market pretty quickly. It’s also not surprising that we have seen a high degree of automated trading develop in Treasury futures, as with our futures markets generally. In recent years, automated trading has accounted for about 67 percent of 10-year Treasury futures, and 64 percent of Eurodollar futures. This is similar, but generally higher than many other futures contracts. For instance, over the last few years, automated traders are on at least one side of 50 percent of trades for metals and energy futures, and almost 40 percent in agricultural contracts.
Electronic trading has contributed to a substantial improvement in transparency in the markets. When you traded in the pit, you could only see the top of the order book – the quotes being shouted out. In the Treasury futures market today, you can see the aggregated trading interest for the top 10 bids and offers.
As regulators, we receive a transaction audit trail that identifies market participants – and a summary of positions held by the largest among them. The transactions data allows us to look at trading by the millisecond. We do not routinely receive all message data, though we can obtain that as necessary, as we did to analyze the events of October 15.
The daily reports we receive are critical to our work, especially our surveillance. Moreover, they were essential in analyzing the events on October 15. Within a few days of the event, this data enabled us to make a general assessment of the activity pattern – including:
And our initial assessment was supported by the later analysis and findings of the Joint Report.
On the whole, this transparency helps us oversee not only whether our markets are fair and robust – but also whether the activities of individual participants are in compliance with our rules. We bring enforcement actions against market participants who engage in fraudulent, manipulative or other illegal actions.
The exchanges also play an important role in market surveillance, as designated self-regulatory organizations.
Flash Events Generally
So what’s the relevance of these structural characteristics in thinking about the significance of what happened on October 15? Well, it’s useful to look at whether this type of event is happening in other products, so let me turn to my second topic. While the Treasury futures market is special in some ways, the structural characteristics of it that I’ve described are shared across all products. There are other key contracts based on well-known benchmarks that have very strong trading volumes, though perhaps not as large as in Treasury futures.
These include a variety of futures on physical products, such as crude and gold for example, as well as financial products – such as the E-mini S&P and EuroFX. As with Treasury futures, all trading is on a single exchange, almost solely in a single, centralized, electronic limit order book, where the trading is highly automated.
So do we see instances of sudden price movements that are largely reversed within a short period of time in other products? If so, are they common? Are they becoming more common?
Our analysis is still quite preliminary, but let me share what we have learned to date.
Our staff measured the frequency of “flash” events in Treasury futures and five of the most active contracts: Corn, gold, WTI crude oil, E-mini S&P, and EuroFX. Each represented the largest volume futures contracts in their respective sectors. The first chart shows the respective volumes. We used a somewhat arbitrary definition of flash events: episodes in which the price of a contract moved at least 200 basis points within a trading hour— but returned to within 75 basis points of the original or starting price within that same hour. This roughly corresponds to the October 15th Treasury market event, where the movement was 210 basis points within one hour and a net change of 60 basis points. We went back to 2010, when we first started receiving a complete daily tape of all transactions.
In just this year, for example, there were about 35 events meeting this definition involving the WTI crude oil contract alone. We also found quite a few having to do with other contracts in the last several years, including corn and gold. The second chart shows the number of these events for various contracts. Movements of a magnitude similar to Treasuries on October 15th were not uncommon in many of these contracts. In fact corn, the largest grain futures market, averaged more than five such events per year over the last five years.
We did a less granular analysis for years prior to 2010 –looking at daily high and low prices to determine whether there were similar types of movements over the course of a day. Daily movements are obviously less significant than hourly movements, and we used a slightly-more restrictive definition of an event: If intraday price movement was more than 350 points, and If the day-over-day price change was less than one-third the daily price range. We examined the number of events per year for some of the larger physical futures contracts, since they have longer histories. We also show the VIX, since that tends to have high volatility, and provides a contrasting example from a different kind of instrument.
So there appear to be more round-trip events in recent years. However, markets were generally volatile in 2008 and 2009. In addition, some of this increase is related to the VIX, which was not significant by volume in earlier years.
We haven’t tried to determine what might have triggered these events. I welcome such analysis, but I think it’s difficult to do. In some cases there might be an obvious cause, but I suspect there are many times where there is not.
We know that markets often move with no apparent cause. John Maynard Keynes said it best: the stock market is a beauty contest in which people are not picking out the prettiest face – but rather the face that they think everyone else believes is most beautiful.
Today, in the futures markets, that determination – regardless of the factors driving it – is being made faster, updated more frequently, and carried out by many people. Every market participant—every algorithm -- can see the order book. Based on analyzing vast quantities of data, every algorithm can react to what every other algorithm is doing – and even what it predicts those other algos will do.
We also know that as with humans, the modern algorithms have risk management capabilities embedded within them. So when there is a moment of sudden, unexpected volatility, it may not be surprising that some in the market pull back – potentially faster than a human can.
The report describes how on October 15, some algos pulled back by widening their spreads and others reduced the size of their trading interest. Whether such dynamics can further increase volatility in an already volatile period is a question worth asking, but a difficult one to answer. It is also very difficult for individual institutions of any type to remain in the book, opposing price headwinds, or worse, to try and catch the proverbial falling knife. For many, this decision can be the difference between risk mitigation and significant losses. Contrary to what some have suggested, I suspect it was difficult for market makers in the pre-electronic era to routinely maintain tight and deep spreads during volatile conditions. They likely took long coffee breaks.
By noting that these types of events happen in other products, I am not suggesting that we simply shrug our shoulders and conclude this is the new normal. To the contrary – I think this provides a helpful perspective to keep in mind as we analyze and think about these issues of Treasury market structure and evolution.
Let me turn to the third issue I would like to discuss with you, which is what we should do next.
The October 15 report lays out some important first steps, and we support them. I think the recommendations regarding data collection in particular are extremely important.
Having adequate data is a necessary first step to explore issues and options. There are significant differences in data collection between the cash market and the futures market. And therefore some steps need to be taken if we are going to collect the same amount of data on the cash market as we routinely get on the futures market. And if we are going to frequently or routinely analyze message data, it will require additional capabilities.
The report also recommends that we take a closer look at algorithmic – or automated – trading. And we have been doing that. So let me spend a few minutes on what we at the CFTC are doing in this regard.
The joint report discusses the potential risks of automated trading. These include: operational risks, such as malfunctioning or incorrectly deployed algorithms. Transmission risk, because shocks based on erroneous orders can have an impact beyond the market where it is placed. Market integrity risk, due to the potential for spoofing or other forms of manipulation.
The increased use of automated trading also raises fundamental issues concerning market structure. Some would say automated trading has created a more efficient marketplace. They would assert that speed has brought significant benefits in terms of reduced transaction costs, improved price discovery, greater liquidity and greater linkages across markets. But others ask whether the speed is worth it?
Can our markets manage it? How should we think about the impact of automated trading on market efficiency, fairness and systemic risk? And what, if anything, should we do about it?
I think all these questions deserve consideration, and we at the CFTC look forward to taking part in that discussion. In the near term, we are focused on looking at operational risks, and taking steps to minimize the potential for disruptions and other operational problems that may arise. And I expect that we will make some new proposals in this regard next month.
By way of background, we have already taken a number of steps to respond to the development of automated trading in our markets, as have the exchanges. A few years ago, we required exchanges to establish risk control mechanisms to prevent market disruptions, including mechanisms that pause or halt trading. And as Chair White noted yesterday, futures exchanges and equity markets have worked to coordinate these mechanisms. We required clearing members to establish risk-based limits for all accounts based on factors such as position size or order size. We also required automatic screening of orders for compliance with risk limits if they are automatically executed.
And we have new statutory authority to prevent new forms of illegal behavior, such as spoofing. We have brought enforcement actions against those who engage in spoofing. And we will continue to be aggressive on this front.
The focus of our forthcoming proposals will be on the automation of order origination, transmission and execution – and the risks that may arise from such activity. These risks can come about due to malfunctioning algorithms, inadequate testing of algos, errors and similar problems. We are concerned about the potential for disruptive events and whether there are adequate measures to ensure effective compliance with risk controls and other requirements.
Now of course, you could have errors before, in the days of pit traders and specialists. You could have failures of systems in less sophisticated times. But generally the consequences were of lesser magnitude than what we may face today. And that’s in large part because the errors were easier to identify, arrest or cure before they caused widespread damage.
I expect that our proposals will include requirements for pre-trade risk controls and other measures with respect to automated trading. These will apply regardless of whether the automated trading is high or low frequency. We will not attempt to define high-frequency trading specifically. I expect that we will propose controls at the exchange level, and also at the clearing member and trading firm level.
I anticipate that our proposals for risk controls will be largely consistent with the best practices followed by many firms already, and they will build on what the exchanges have already done. While we would, in some cases, propose the types of controls necessary, we will not prescribe the parameters or limits of such controls.
For example, we will likely propose requiring pre-trade controls, such as message throttles and maximum order size limits. But we will not prescribe how those limits should be set. We are looking at proposing requirements pertaining to the design, testing and supervision of automated trading systems – as well as measures such as “kill switches,” which facilitate emergency intervention in the case of malfunctioning algorithms.
Further, we are looking at whether to require registration for proprietary firms not already registered with the CFTC who access the market directly and who are also using automated trading – to facilitate more effective oversight.
And we are also looking at whether to require measures to limit the practice of self-trading. I would also distinguish unintentional self-trading from wash trading, which is illegal. We are also looking at market maker and trading incentive programs, which have become more significant as automated trading has increased. I expect our focus here will be on increased transparency.
Let me make it clear that the Commission has not yet decided to issue any proposals, so my comments reflect my own views. And of course, any proposals will be subject to public comment. And we always look forward to your input.
Cybersecurity and Technological Risk Generally
I also want to briefly mention one other related area of concern, and that is cybersecurity. In today’s highly electronic and automated markets, the disruptions that can be caused by automated trading are one type of operational or technological risk. The risk of cyberattacks by those looking to profit – or intentionally disrupt – our markets is another.
While we have been addressing this issue through our regulations and examinations, we are also considering some additional proposals. These would focus on making sure the private companies that run the core infrastructure under our jurisdiction – such as the major exchanges, clearinghouses, and swap data repositories – are doing adequate evaluation of these risks and testing of their own cybersecurity and operational risk protections.
As with automated trading, I expect these will be principles-based standards. For example, our proposals will describe the general types of testing that are needed—including controls testing, vulnerability testing and penetration testing –but we will leave the detail of how to do the testing to the responsible firms.
As with automated trading, the Commission has not yet made a decision so the views I have expressed are my own. Any proposals that we do make will also be subject to public comment.
I want to conclude with just one final thought.
A lot of views have been exchanged at this conference. And I look forward to continuing the discussion in the weeks and months ahead.
I would simply suggest that we should take our time in drawing conclusions. We should resist the temptation to recognize an effect and presume the cause.
And we should make sure our analysis is data-driven. Good, data-driven analysis, such as the October 15 report, takes resources. There’s more we should be doing to understand the changes in the markets. We should be looking at more data and more information, more quickly.
But the CFTC’s current budget has simply not kept up with the growth of the markets. We need to stay on top of technological innovation and growth. Congress has responded with modest budget increases – which we appreciate. But without additional resources, it is difficult for us to carry out our mission in a manner that the American people deserve.
I look forward to continue working with everyone in this room to achieve those goals.
Last Updated: October 21, 2015